Persistency

Based on operational limits, persistency analysis determines the extent and the occurrence of sufficiently long windows from the 3-hourly time series at the model output point (displayed in the header of the page). A window is a continuous time interval of specified minimum duration and with specified conditions. The analysis is capable of determining continuous periods of favourable or unfavourable conditions. For each month of the year, the analysis results in

• The percentage of the time covered by sufficiently long windows
• The number of sufficiently long windows
• The delay waiting for sufficiently long windows

Method

Per month of the year the above quantities are found for all years analysed as well as for each year separately (inter-annual variation). Windows crossing month boundaries are counted for a particular month if the part of the window covering that particular month is at least as long as the minimum window required. If both parts (on both sides of a month crossing) of a sufficiently long window are shorter than the minimum window required, the window is only counted for the month covered by the longer part of the window. The delay waiting for specified conditions is found as the mean duration between all model-dates in the month of interest (model-dates are normally three hours apart) and the start of the next sufficiently long window in the current month or in the next month. The analysis searches for the start of the next available window from the initial date and is allowed to continue into and up to the end of the next month. If no suitable window is found within this timeframe then the start of the window is set to the start of the following month i.e. the start of the second month from the initial date. For example, if neither July (31 days) nor August (31 days) contain the start of a window, the delay for July will be 46.5625 days: the mean of the values 62 (delay in days between July 1, 00:00 UTC and September 1, 00:00 UTC), 61.875 (delay in days between July 1, 03:00 UTC and September 1, 00:00 UTC), ... , 31.125 (delay in days between July 31, 21:00 UTC and September 1, 00:00 UTC). Note that even if there are no windows at all (an infinite delay), the above method will still yield a lower limit.

Remarks

• We update the waveclimate.com database once a year: in spring we add the model data and satellite data of the year before. This means that the data ends at e.g. December 31 2010 at 21:00 for a model time step of 3 hours. This affects the persistency analysis in the following way. In case all sea states are approved (no limits in effect), the percentage of time covered by windows should be 100 for each month. However, this is not exactly true for the last December month analyzed: the fraction of time covered by windows in December of the last year analyzed is 100*(1-3/(31*24))=99.60 percent. Also be aware of the fact that the mean delay for the month December of the last year analyzed might deviate from the delays in the month December of the other years due to the fact that the end of the database is reached: there is no next month. This effect becomes most apparent if very few or no windows are found.
• In general, persitency analysis is used to estimate workability and upper limits are applied to one or more wind/wave parameters. We estimate workable time (conservatively) as the interval between the first and the last of a series of consecutive workable sea states. This is best explained by an example:
  An operation might require waves below 1m. Suppose that there is only 1 model record from the 3-hourly model series for which wave height remains below this limit. In this case, no workable window will be found. If there are two adjacent records with waves below 1m, a weather window of 3 hours would be determined. A weather window of 6 hours requires three consecutive samples of wave height below 1m, and so on.

Inputs

The persistency screen asks you to select the following

• The type of output (fraction of time covered by windows, number of windows, delay waiting for windows)
• The type of wave period used (conditions can be set for peak, mean or zero-crossing wave period)
• The set of wind and wave parameters to which conditions will be applied simultaneously
• The type of conditions applied (interpret limits as lower or upper limits)
• The limits to be applied to the wind and wave parameters

The type of output (radio buttons)

Persistency output can be expressed in terms of percentage of time, number of windows, or delay. All persistency output pages contain one table and one plot. The table shows the selected quantity per month (columns) and per year (rows), including the monthly mean taken over all years (addtional row). The plot depicts monthly means and their interannual variation.

The type of wave period used (radio buttons)

Conditions for wave period, if any, will be applied to peak wave period, mean wave period or zero-crossing wave period according to your choice.

The set of wind and wave parameters (check boxes)

Conditions may be applied simultaneously to one or more wave/wind parameters in each model record (seastate). Wave height and wave period can be based on the total wave spectrum, the swell part of the spectrum or the wind-sea part. For workability analysis for example, 1m high swell will probably cause more problems than wind waves of 1m high. You may combine limits for wave height and wave period to set your own definition of swell or wind-sea, e.g. the total wave height should remain below 0.5m if the corresponding wave period exceeds 8s while waves of up to 1m are allowed as long as the corresponding wave period is less than 8s.

The type of conditions applied (radio buttons)

This input allows you to search the model series for values below or, alternatively, above a specified limit. In general, persistency analysis is used to determine workability and constraints will be imposed: limits are not to be exceeded. Windows are periods of calm weather in this case. One could also be interested however in the persistency of extreme values and apply limits as a threshold. Now, windows are periods of excursion above the specified limit.

The limits to be applied (text fields)

The set of limits entered here will be applied simultaneously to the corresponding wave and wind parameters. So for example, when given an upper limit of 1m for total wave height and an upper limit of 5m/s for wind speed, the server will search for sufficiently long weather windows during which for each sea state, i.e. for each 3-hourly model record of wind and wave parameters, both wave height and wind speed remain below the specified limits.

Mind that any limit to be applied to wind speed is expressed in terms of m/s or knots, according to your preferences.

The minimum window length (text field)

Here you specify the minimum window length required expressed as hours. Windows cannot be shorter than the output time step of the model (3 hours) and the value entered here will be rounded to the nearest multiple of the output time step of the model.

Tip: Logon as demo (no password needed) to get acquainted with the different types of output without being charged.

Tip: Use preferences to tailor the output to your needs (under Account in the upper right menu) .